The DNS problems have resurfaced again today. Amazingly, the only news story on this is still Dean Pullen in The Inquirer with a follow-up article (surely this is a major story? Where are the other news sites?). But, perhaps because of Twitter and Dean Pullen's article, the good news is that Register.com is now itself using Twitter to let the world know that it is working on the problem. Good transparent customer service.
This makes 3 days in a row with major DNS outages. An important new consequence of DNS outages is the Cloud Computing Angle. Here at Vordel, our customers can use our XML Gateway to connect up to Web Services exposed by Cloud providers (e.g. SalesForce and Amazon). Like a local "cable box" for connecting to the Cloud. Those connections to services such as Amazon S3 for storage and Force.com for CRM are important for their businesses, so it makes sense to put the connections up to the “Cloud” services through a Vordel Gateway, which monitors the connection. If an outage happens, then the Vordel box takes remedial action by sending a notification, using its cache, and rerouting the traffic. Yesterday and today's Register.com outage didn’t affect Cloud traffic, but the earlier UltraDNS attack did. So, if you’re using Cloud services, then that means that your operations now depend not only on the Cloud service provider, but on the DNS system. It makes sense to put intelligence at the client side which is connecting up to the Cloud.
So, you have to have that local monitoring in place or else a DNS attack knocks out your own core services which now depend on Cloud services. It is like the old definition of a Distributed System – one in which a component you’ve never previously heard of can bring the whole thing down.
0 comments:
Post a Comment